You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@kafka.apache.org by "Ewen Cheslack-Postava (JIRA)" <ji...@apache.org> on 2014/12/05 03:22:12 UTC

[jira] [Commented] (KAFKA-1807) Improve accuracy of ProducerPerformance target throughput

    [ https://issues.apache.org/jira/browse/KAFKA-1807?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14234940#comment-14234940 ] 

Ewen Cheslack-Postava commented on KAFKA-1807:
----------------------------------------------

Created reviewboard https://reviews.apache.org/r/28735/diff/
 against branch origin/trunk

> Improve accuracy of ProducerPerformance target throughput
> ---------------------------------------------------------
>
>                 Key: KAFKA-1807
>                 URL: https://issues.apache.org/jira/browse/KAFKA-1807
>             Project: Kafka
>          Issue Type: Improvement
>          Components: clients, tools
>    Affects Versions: 0.8.1.1
>            Reporter: Ewen Cheslack-Postava
>            Assignee: Ewen Cheslack-Postava
>            Priority: Minor
>             Fix For: 0.8.3
>
>         Attachments: KAFKA-1807.patch
>
>
> The code in ProducerPerformance that tries to match a target throughput is very inaccurate because it doesn't account for time spent sending messages. Since we have to get the current time to timestamp the messages, we can be much more accurate by computing the current rate over the entire run and only add to the sleep deficit if we're above the target rate.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)